15:01:11 <carl_baldwin> #startmeeting neutron_l3
15:01:11 <openstack> Meeting started Thu Apr 10 15:01:11 2014 UTC and is due to finish in 60 minutes.  The chair is carl_baldwin. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:12 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:14 <openstack> The meeting name has been set to 'neutron_l3'
15:01:38 <carl_baldwin> #topic Announcements
15:01:49 <carl_baldwin> #link https://wiki.openstack.org/wiki/Meetings/Neutron-L3-Subteam
15:02:06 <xuhanp> hi
15:02:14 <otherwiseguy> hi
15:02:33 <carl_baldwin> Deadline to submit summit design topics is April 20th.  About a week away.  Is anyone planning to submit a topic (or already submitted)?
15:03:27 <carl_baldwin> I'll probably have some time to read through some if you want an extra set of eyes on it before submitting.
15:03:54 <carl_baldwin> I'm busy myself trying to get a couple of blueprints written up to submit.
15:04:47 <carl_baldwin> Does anyone have the link for the page for submitting these handy?
15:05:04 <overlayer> http://summit.openstack.org/cfp/create
15:05:48 <carl_baldwin> overlayer: Thanks.
15:06:22 <carl_baldwin> #topic l3-high-availability
15:06:42 <carl_baldwin> safchain: any updates?  I saw some update to the blueprint.  Accepted for Juno.
15:07:13 <safchain> carl_baldwin, hi, just got some interesting comments
15:07:39 <safchain> carl_baldwin, I have to check if there is overlaps with le l3 dvr patch
15:08:23 <carl_baldwin> Yes, working closely with DVR will be a good thing.
15:08:30 <safchain> carl_baldwin, I will ping swami to talk about that
15:08:57 <safchain> carl_baldwin, that's all for me
15:09:04 <carl_baldwin> Swami will join a bit later.
15:09:42 <carl_baldwin> For me, reviewing this has fallen below some Icehouse stuff and the summit topics.  I plan to get back to it after the 20th.  I imagine others are in the same boat.
15:09:53 <carl_baldwin> Thanks for your update.
15:10:09 <carl_baldwin> #topic bgp-dynamic-routing
15:10:11 <safchain> carl_baldwin, sure, np
15:10:17 <carl_baldwin> nextone92: hi
15:10:24 <nextone92> Hi carl!
15:10:39 <carl_baldwin> I had an action to update the use cases.  I have done that and they are linked from the sub team page.
15:10:44 <nextone92> Unfortunately I haven't gotten any additional feedback about the document
15:11:27 <carl_baldwin> I did have another look and I like the changes you've made to it.
15:11:47 <nextone92> Thank you Carl
15:11:51 <carl_baldwin> Do you plan to suggest a summit session?
15:12:06 <nextone92> Would that be okay?
15:12:50 <carl_baldwin> Yes, I think it would be great.  I'd be happy to discuss it sometime during the next week.
15:13:21 <nextone92> Thank you! I will submit the proposal
15:13:40 <nextone92> And looking forward to any additional comments and feedback
15:13:59 <carl_baldwin> Great.  Feel free to ping me on IRC.
15:14:02 <carl_baldwin> Anything else?
15:14:16 <nextone92> That's all on dynamic routing
15:14:36 <carl_baldwin> Thanks.
15:14:51 <carl_baldwin> #topic rootwrap-daemon-mode
15:15:01 <carl_baldwin> ajo, YorikSar : hi
15:15:17 <YorikSar> o/
15:15:20 <carl_baldwin> I've been out of the loop on these reviews this week.
15:15:32 <YorikSar> I don't have much updates this week.
15:15:59 <carl_baldwin> Okay, let's talk more about it after the 20th.
15:16:20 <YorikSar> I've ran into unpleasant bug in stdlib that isn't fixed on 12.04 that's used in gate but was fixed in the very next version of Python (2.7.4)
15:17:17 <YorikSar> Ok
15:17:26 <carl_baldwin> YorikSar: Could we bring it up with infra to see if an upgrade is possible?
15:17:59 <YorikSar> carl_baldwin: I suppose we should support vanilla Ubuntu 12.04
15:18:19 <YorikSar> And I don't think it's possible to upgrade Python in Ubuntu LTS release
15:18:21 <ajo> hi carl_baldwin , YorikSar , sorry, it seems that I was a little bit confused with the meeting time , and time shifts
15:18:30 <ajo> (reading)
15:18:42 <carl_baldwin> ajo: Thanks for joining.
15:19:00 <ajo> YorikSar,
15:19:03 <ajo> what's the bug exactly?
15:19:07 <YorikSar> But I can ask on #openstack-infra about it.
15:19:16 <YorikSar> http://bugs.python.org/issue6065
15:19:19 <ajo> I mean, in RH they use py2.6
15:19:27 <ajo> so I'll need to make sure this is backported in place
15:19:42 <YorikSar> Oh...
15:19:49 <carl_baldwin> YorikSar: It might help to understand.  The worst that can happen is they will tell us we can't upgrade.
15:19:54 <carl_baldwin> Is the bug also in 2.6?
15:20:00 <YorikSar> I'm trying to come up with workaround for that.
15:20:14 <ajo> are you sure it's http://bugs.python.org/issue6065 ?
15:20:21 <ajo> it looks like MSI installer related
15:21:02 <YorikSar> Hm... I have no idea about 2.6 I guess it hasn't been fixed in that branch since it was already not supported by then.
15:21:16 <YorikSar> Oh... I've lost some digit. Let me check.
15:22:00 <carl_baldwin> YorikSar: ajo: I'll let you take it offline.  Looks like you've got some things to try.
15:22:28 <ajo> thanks carl_baldwin
15:22:36 <carl_baldwin> #topic Multiple Subnets on External Network
15:22:46 <YorikSar> carl_baldwin: ok
15:22:55 <carl_baldwin> This is a new topic on the agenda this week.
15:23:30 <carl_baldwin> I've heard this from a few people now and it is a problem that I have as well.
15:24:32 <YorikSar> It was http://bugs.python.org/issue6056
15:24:49 <YorikSar> (sorry for interrupting)
15:25:03 <carl_baldwin> YorikSar: (no worries)
15:25:30 <carl_baldwin> The use case for this is to allow growing the floating IP pool by adding subnets to an external network.
15:26:43 <carl_baldwin> I don't have much more than that, just an introduction to the topic.  Please contact me if this use case is interesting to you and you would like to participate in making it work like it should.
15:27:34 <carl_baldwin> I hope to understand it a little better next week.
15:27:54 <xuhanp> carl_baldwin, I am interested in it a lot :-) any information to learn more about this?
15:28:11 <safchain> carl_baldwin, sure, I'm interested too
15:29:00 <Swami> carl: hi
15:29:03 <carl_baldwin> xuhanp: Not really yet.  I'll keep you in the loop.
15:29:08 <carl_baldwin> safchain: you too.
15:29:12 <xuhanp> carl_baldwin, thanks
15:29:16 <safchain> carl_baldwin, thanks
15:29:25 <safchain> Swami, hi
15:29:30 <carl_baldwin> If you get ambitious and start looking in to it, do the same for me.  :)
15:29:31 <Swami> safchain: hi
15:29:51 <carl_baldwin> Swami: your timing is impeccable
15:30:08 <carl_baldwin> #topic neutron-ovs-dvr
15:30:23 <Swami> DVR progress update
15:30:54 <Swami> As mentioned in the previous meeting we could not post the L2 Agent and L3 Agent code this week, since we were busy with some merging work.
15:31:24 <Swami> The plugin code comments have been addressed and patch5 is up for review.
15:31:36 <Swami> Still working on some more unit testing before removing the WIP.
15:32:01 <Swami> We are targeting to push the L2 Agent and L3 Agent code by next week.
15:32:35 <Swami> Meanwhile in yesterday's discussion there was question that was raised about the multiple external networks and using the br-int instead of the br-ex.
15:32:56 <Swami> I am not sure if there was a discussion on this with safchain in this meeting today.
15:33:07 <safchain> Swami, not yet
15:33:29 <carl_baldwin> That will be fine.  I imagine that the community will have more time for reviewing as Icehouse winds down and after the deadline for submitting summit sessions on the 20th.
15:33:45 <Swami> safchain: Today in Icehouse when we use multiple external networks, the networks are bridged from the br-int and not from the br-ex.
15:33:54 <safchain> Swami, that was the subject of the email I sent to you
15:34:15 <Swami> safchain: Yes I did not understand the complete context when you sent me the email.
15:34:17 <safchain> Swami, yes with provider networks and bridges_mappings
15:34:24 <Swami> sorry it was my mistake.
15:34:28 <safchain> Swami, np
15:35:04 <Swami> So when provider networks are in place where is the actual SNAT happening.
15:35:12 <Swami> Is it in the network node or in the provider network.
15:35:20 <safchain> Swami, so yes with multiple networks there is a veth between br-int and the external bridge
15:35:54 <safchain> in the network node, between the qg- and the qr
15:36:07 <safchain> but the qg is in the br-int
15:37:13 <Swami> So if the qg is in the br-int then what is the use of the veth pair between the external bridge and the br-int
15:38:03 <safchain> so today it is a veth but with this patch https://review.openstack.org/#/c/76685/, it will use a patch interface
15:38:35 <safchain> Swami, map the local vlan to the ext vlan
15:39:26 <Swami> safchain: So the external traffic will hit the br-int qg_ and will leave the network node without touching the br-ex
15:39:38 <safchain> right, no more br-ex
15:40:23 <Swami> In this case will the existing services such as VPN will it still work.
15:40:44 <xuhanp> safchain, but the external bridge is still needed just like other physical interface bridge, right?
15:40:47 <safchain> Swami, but you can use provider network for a service network private or shared
15:41:09 <safchain> xuhanp, no not needed
15:41:23 <safchain> xuhanp, you can define a provider network as external
15:41:37 <Swami> safchain: Ok got it.
15:41:45 <xuhanp> safchain, OK. I see
15:41:50 <Swami> In this case all the services will then be rendered by the provider network.
15:42:12 <safchain> we have a lot of customers which use provider networks
15:42:13 <Swami> Does the provider network also takes care of assigning the Floating ip
15:42:46 <safchain> Swami, it is exactly the same as with the br-ex
15:43:25 <Swami> safchain: got it.
15:43:36 <safchain> Swami, it is just the ovs lib which doesn't create the interface in the same way thanks to the bridge_mappings config parameter
15:43:56 <Swami> If I have more questions I will ping you on this, while we are working on the North-South DVR implementation.
15:44:12 <safchain> Swami, but you could have any external network, so any ext bridge
15:44:38 <safchain> Swami, I would like to talk with you about the l3 patch you submit and the vrrp
15:44:59 <safchain> I don't have the time today, will you available tomorrow
15:45:10 <Swami> safchain: Feel free to shoot me an email.
15:45:19 <safchain> Swami, great, i'll do
15:45:30 <Swami> I will be available tomorrow. If you would like to chat on IRC send me an email and I can log in.
15:46:00 <safchain> Swami, ok, and we could talk about the ext net. as well
15:46:25 <carl_baldwin> safchain: Swami: Is there a summary of what needs to happen wrt br-int/br-ex in the context of DVR or VRRP?
15:47:00 <Swami> carl: Yes the DVR team need to evaluate the Ovs rules for the provider network case and validate the scenario.
15:47:11 <Swami> I will probably give an update in the next week about this.
15:47:30 <safchain> carl_baldwin, I think for the VRRP part it should be ok
15:47:43 <carl_baldwin> Swami: safchain:  Great, thank you.
15:47:51 <Swami> In the context of the DVR - L3 vrrp patch I will work with safchain and see how we can bring it together.
15:47:54 <safchain> carl_baldwin, already tested, but of course needs more tests
15:48:46 <Swami> carl: That's all I have from the DVR team.
15:48:55 <carl_baldwin> Swami: Thanks for the update.
15:48:58 <safchain> sorry guys I have to go now, Swami see you tomorrow, carl_baldwin thx for the meeting
15:49:12 <carl_baldwin> safchain: bye
15:49:17 <safchain> bye
15:49:19 <Swami> safchain: thank you
15:49:27 <carl_baldwin> #topic Open Discussion
15:50:58 <overlayer> safchain, cya
15:51:39 <carl_baldwin> Thanks to all who have participated.  Have a great week!
15:51:54 <Swami> bye
15:52:02 <carl_baldwin> #endmeeting